The borderline diagnosis enjoys a widespread and growing popularity among clinicians. In some hospital settings, up to twenty-five percent of admisions receive this diagnosis. Borderline patients are among the most difficult to treat, often presenting a bewildering array of affective and cognitive symptoms, impulsive and self-destructive behaviors. Despite the magnitude and severity of the clinical problem presented by these patients, no systematic controlled drug tretment studies have ever been undertaken. The reason for this striking paucity of research has been the lack of reliable diagnostic criteria and an objective assessment method. The diagnostic criteria for borderlines, developed and tested by Spitzer, Endicott and Gibbon, and the new Diagnostic Interview for Borderlines by John Gunderson, provide an objective and reliable method for defining a sample of borderline patients for a first placebo controlled, double blind drug treatment study. This study will compare the clinical efficacy of amitriptyline and haloperidol in a placebo controlled double blind protocol. Symptom response to drug treatment will be assessed on standard scales with special emphasis given to affective, cognitive and impulsive symptoms. Recently developed assay methods will be used to assure pharmacologic control over systemic exposure to medication. All patients will receive the individual, group and milieu therapies which are standard hospital treatments for borderline patients. In addition to determining the main effects of the two drugs, clinical responses will be analyzed for patients meeting the subtype criteria for Unstable or Schizotypal Personality Disorders (or both) - two statistically independent dimensions of the borderline disorder which co-exist in half of borderline patients. A syndromal profile will be generated for haloperidol and amitriptyline responders to provide a first empirical treatment guide for clinical use. This proposal is a "state of the art" response to a major clinical problem. This is a revision of ADAMHA application MH 35392-01 A1.